Score every action.
Rank up your defense.
One API call blends device signals, behavior, IP intelligence, and a live fraud-ring graph into a 0–100 score across 17 named fraud vectors — carding, promo abuse, BNPL, account takeover, and more. Send more traffic, unlock a cheaper rank.
One score.
Five signal families.
Device telemetry, behavioral biometrics, IP intelligence, relationship-graph insights, and AI contextual review combine into a single fraud score — and a combo multiplier rewards catching multiple red flags at once.
Try the combo multiplier.
Toggle signals to watch the score move. Fire two HIGH/CRITICAL signals together and the accumulated score gets multiplied ×1.2 — the same synergy bonus the live engine applies.
Rooted/jailbroken device, emulator, or app tampering detected
VPN, public proxy, or Tor exit node behind this session
Short graph distance to a confirmed fraud node or ring
Non-human typing velocity or paste-only form completion
One API call.
Declare the fraud vector, send the session, get a decision. An unrecognized action_type or a missing field is rejected before scoring even runs.
const res = await
fetch('/v1/score', {
method: 'POST',
body: JSON.stringify({
session_id: 'sess_8f3k...',
action_type: 'carding_fraud',
user_data: { card_hash, card_bin, billing_country }
})
});
// Response
{ decision: 'BLOCK',
fraud_score: 94,
risk_factors: [...], reason: '...' }Not a generic bucket — 17 named vectors.
Every integration declares exactly which fraud vector it's guarding — each with its own required business fields, cross-checked against device, network, and graph signals on every single request.
Built for the whole flow.
Five steps from trigger
to logged decision.
A contract-first integration: the SDK collects and encrypts, your backend declares the fraud vector, the engine detects and decides — every step of the way is auditable.
Collect & Encrypt
The SDK (iOS, Android, or web) captures a device fingerprint — rooted/emulator flags, sensor variance, canvas hash, GPS — and encrypts it end-to-end with your project's own keypair before it ever leaves the device.
Open Session
The encrypted fingerprint is sent to open a session. FDS returns a session_id your backend reuses for every action that user takes next.
Declare & Score
On checkout, login, a loan application, or any meaningful action, your backend calls the scoring endpoint with the session_id, one of 17 fraud vector codes, and that vector's required fields. Anything malformed is rejected before scoring runs.
Detect
A blocklist match blocks instantly. Otherwise the graph and IP-intelligence lookups run in parallel (fail-open if either is slow), ~50 weighted rules produce a running score, and genuinely ambiguous mid-range scores get an extra AI contextual review under a strict time budget.
Decide & Log
You get a decision (APPROVE / REVIEW / BLOCK), the risk factors that fired, and a plain-language reason — plus a flag for whether the decision used complete data. Every decision is written to the audit incident log in the background.
Send more traffic.
Pay less per event.
Prepaid, per-event pricing in IDR — no subscription, no contract. Your trailing 30-day volume decides your rank; every rank up drops your rate, for high-value and low-value events alike.
Every new project starts with an automatic onboarding bonus — no card required. That's enough for ~1,500 fraud checks at Bronze tier before you ever top up.
FAQ
What counts as a scored request?
Every call to the scoring endpoint counts as one event, priced by your project's current 30-day volume tier and whether the fraud vector is high-value (payment-adjacent) or low-value (identity/session).
Can a merchant under-declare an event to pay less?
No — if the business data is clearly payment-shaped (a card hash, voucher code, purchase amount, or disbursement account), the event is force-classified high-value regardless of the declared action_type.
Is my usage tier enforced automatically today?
Tier pricing, your prepaid balance, and every billing record are fully tracked and visible in the console right now. Automatic balance deduction and hard cutoff at the scoring endpoint are on our roadmap, not live yet.
Everything you'd
ask your fraud ops team.
Scoring is the API. Understanding why it decided what it decided lives here — one console per merchant, scoped per project, with your rank and billing always in view.
One merchant account, many projects — each with its own API key, encryption keypair, prepaid balance, and blocklist. Nothing crosses project lines.
Add a bad IP, email, or phone number once. Every future match is an instant block, skipping the rest of the pipeline entirely.
Every scored decision — request and response — is searchable history, so fraud ops can reconstruct exactly why any call resolved the way it did.
Volume trend, fraud rate, false-positive rate, and which rules are firing most — streamed in real time to the console, not a monthly export.
Start at Bronze.
Rank up from there.
Every new project opens with a Rp 50,000 onboarding bonus — enough for ~1,500 fraud checks at Bronze tier. No card, no contract, no subscription. Just a scoring endpoint and a ladder to climb.